Table 4 Performance of all the predictors on secondary-structure motifs on the test set TS1.

From: RNA secondary structure prediction using an ensemble of two-dimensional deep neural networks and transfer learning

 

Stem(F1a)

Stem (PR)

Stem (SN)

Hairpin loop (F1a)

Hairpin loop (PR)

Hairpin loop (SN)

Bulge (F1a)

Bulge (PR)

Bulge (SN)

Internal loop (F1a)

Internal loop (PR)

Internal loop (SN)

Multiloop (F1a)

Multiloop (PR)

Multiloop (SN)

SPOT-RNA

0.762

0.841

0.697

0.686

0.625

0.761

0.369

0.508

0.289

0.266

0.239

0.300

0.562

0.503

0.638

mxfold

0.717

0.769

0.671

0.625

0.525

0.771

0.213

0.360

0.152

0.329

0.270

0.422

0.526

0.465

0.607

ContextFold

0.706

0.755

0.663

0.633

0.513

0.825

0.286

0.539

0.194

0.214

0.170

0.289

0.574

0.544

0.607

CONTRAfold

0.688

0.705

0.671

0.624

0.553

0.715

0.331

0.378

0.294

0.279

0.241

0.331

0.469

0.587

0.391

Knotty

0.670

0.739

0.613

0.600

0.493

0.766

0.295

0.421

0.227

0.279

0.238

0.338

0.549

0.649

0.476

IPknot

0.665

0.754

0.595

0.602

0.510

0.735

0.201

0.474

0.128

0.218

0.202

0.236

0.417

0.339

0.542

RNAfold

0.671

0.686

0.657

0.617

0.539

0.722

0.313

0.500

0.227

0.270

0.218

0.354

0.514

0.555

0.478

ProbKnot

0.625

0.661

0.592

0.571

0.480

0.704

0.276

0.377

0.218

0.209

0.187

0.236

0.481

0.492

0.470

CentroidFold

0.646

0.662

0.632

0.579

0.467

0.761

0.293

0.395

0.232

0.179

0.211

0.156

0.433

0.379

0.506

RNAstructure

0.646

0.665

0.629

0.596

0.508

0.720

0.300

0.440

0.227

0.238

0.204

0.285

0.478

0.546

0.424

RNAshapes

0.627

0.650

0.605

0.574

0.507

0.663

0.310

0.432

0.242

0.238

0.193

0.308

0.433

0.507

0.378

pkiss

0.618

0.684

0.565

0.532

0.449

0.655

0.253

0.457

0.175

0.229

0.183

0.304

0.406

0.494

0.344

CycleFold

0.496

0.431

0.584

0.437

0.564

0.357

0.277

0.333

0.237

0.000

0.000

0.000

0.367

0.374

0.360

  1. \({}^{a}\)Harmonic mean of precision (PR) and sensitivity (SN)